Increasing productivity and quality of care: Robot-aided ...KREBS et al. Robot-aided neuro-rehabilitation r World Health Organization (WHO) is predicting that the population over 65

in

Department ofVeterans Affairs

Journal of Rehabilitation Research andDevelopment Vol . 37 No . 6, November/December 2000Pages 639-652

Increasing productivity and quality of care : Robot-aidedneuro-rehabilitation

H.I. Krebs, PhD; B.T. Volpe, MD ; M.L. Aisen, MD; N. Hogan, PhDMassachusetts Institute of Technology, Mechanical Engineering Department, Newman Laboratory for Biomechanics andHuman Rehabilitation, Cambridge, MA 02139; Massachusetts Institute of Technology, Brain and Cognitive Sciences

Department, Cambridge, MA 02139; Cornell University Medical College, Department of Neurology and Neuroscience,

Burke Institute of Medical Research, 1300 York Avenue, New York, NY 10021 ; Veterans Health Administration,

Department of Rehabilitation and Development, 810 Vermont Avenue, N .W. (122), Washington, DC 20420

Abstract—This paper presents an overview of our research inrobot-aided stroke neuro-rehabilitation and recovery . At theonset of this research we had to confront squarely (and solve!)a critical question : If anatomy is destiny, can we influence it?Our efforts over the last five years have been focused onanswering this question and we will present a few of our clini-cal results from over 2,000 hours of robot-aided therapy with76 stroke patients . To determine if exercise therapy influencesplasticity and recovery of the brain following a stroke, weneeded the appropriate "microscope" that would allow us toconcomitantly control the amount of therapy delivered to apatient, while objectively measuring patient's performance.Back-driveable robots are the key enabling technology. Ourresults to date using common clinical scales suggest that robot-aided sensorimotor training does have a genuinely positiveeffect on reduction of impairment and the reorganization of theadult brain. Yet while clinical scales can help us to examine theimpact in the neuro-recovery process, their coarse naturerequires extensive and time-consuming trials, and on top of thatthey fail to show us details important for optimizing therapy.Alternative, robot-based scales offer the potential benefit of

This work was supported in part by The Burke Medical ResearchInstitute, NSF under Grant 8914032-BCS, and NIH under Grant R01-HD37397-01 and Grant R01-HD36827-02.Address all correspondence and requests for reprints to : H .I. Krebs, 77Massachusetts Avenue, 3-137, Cambridge, MA 02139-4307 ; email:[email protected].

new finer measurements and deeper insight into the processof recovery from neurological injury. We also plan to use pre-sent technology to establish the practicality and economic fea-sibility of clinician-supervised, robot-administered therapy,including classroom therapy. We feel quite optimistic that themarch of progress will accelerate substantially in the nearfuture and allow us to transfer this technology from theresearch realm to the everyday treatment of stroke survivors.

Key words : robot, robot-aided neuro-rehabilitation, robot-aided stroke rehabilitation, stroke, stroke rehabilitation.

INTRODUCTION

During his frequent testimony before Congress, theFederal Reserve Board Chairman, Alan Greenspan,attributed the longest period of economic expansion inAmerican history to the information technology that isreshaping America. Unemployment rates are at a recordlow of a mere 4 percent, while inflationary pressure isclearly controlled, at less than 3 percent . This unheard-ofgrowth and prosperity has puzzled more than a few econ-omists and fortunetellers . To explain this anomaly, noth-ing is more appropriate than paraphrasing anothermoneyman, Paul Krugman of MIT: "Productivity isn't

639

640

Journal of Rehabilitation Research and Development Vol . 37 No . 6 2000

everything, but in the long run it is almost everything"(1) . Yet, societal memory typically sells short and we for-get the anguish of years in which many questioned if anyreal increase in productivity would ever occur despitetechnophile employees avidly spending working andnonworking hours entertaining themselves with the newpuzzles offered by computers and networks . It paid off:After enduring their share of pain, "old industries" are re-inventing themselves and productivity is soaring (com-petitiveness is an effect, not a cause).

Productivity has not increased at an equal paceacross different industry sectors . Upheavals within thepresent health care system strongly suggest that it is mori-bund and may be the next "old industry" to undergo mas-sive restructuring . In fact, it may well be the hardest andmost regulated conversion . We cannot afford a temporaryshutdown for restructuring but we may not be able toafford the present system either: The Health CareFinancing Administration (HCFA) projected health carecosts to surpass 16 .6 percent of the total GNP in the year2007 ($2.1 trillion).

This situation creates a pressing need and an oppor-tunity, both of which motivate our research : The need isfor new therapeutic strategies to increase productivitywhile optimizing the quality of care ; the opportunity is totake advantage of recent dramatic advances in technolo-gy—especially in robotics, sensing, information process-ing, and telecommunications . In particular, our researchgoal is to develop innovative treatments that take advan-tage of robotics and information technology to enhanceneuro-rehabilitation . Our approach is a departure frommost prior work using robotics for rehabilitation ; ratherthan developing rehabilitation robots to assist a personwith disabilities, we are creating robot-aids to assist andsupport clinicians in their efforts to facilitate a disabledindividual's functional recovery, while enhancing thehealthcare system productivity.

There are three ways to increase productivity in thedelivery of rehabilitation without sacrificing quality ofpatient care: a) develop evidence-based therapy (for exam-ple, deliver the optimal therapy to the particular patient'sneed), b) re-allocate personnel and tasks (for example, min-imize paperwork, freeing more personnel to deliver care),and c) increase the productivity of each caregiver (forexample, provide therapists with appropriate tools).Robotic aids can impact all these modes and increase pro-ductivity not only by introducing new efficiencies into cer-tain routine physical and occupational therapy activities,

i To our knowledge, Steven Lehman (University of California, Berkeley)but also by providing a rich stream of objective data to

coined the term "rehabilitators ."

assist in patient diagnosis, prognosis, customization of ther-apy, assurance of patient compliance with treatment regi-mens, and maintenance of patient records.

This technology promises to have an impact on abroad range of neurological conditions, encompassing alarge class (if not the entire gamut) of potentially dis-abling conditions . While millions of people in the U .S.acquire movement disabilities as the result of injury anddisease—e.g., stroke CVA, traumatic brain injury (TBI),multiple sclerosis (MS), spinal cord injury (SCI),Parkinsons disease (PD)—it would be unrealistic toattempt to develop and evaluate the technology in all ofthese applications . For that reason we focused our initialefforts on stroke, the leading cause of disability in theU.S. with 700,000 new cases every year.

Notwithstanding the exciting possibilities of robot-ics and information technology, it is imperative to estab-lish from the outset how well these technologies reallywork. Evidence-based healthcare casts a cynical (but sci-entifically appropriate) view on current rehabilitationpractices, perhaps best captured by paraphrasing a remarkattributed to Voltaire: "Physicians, nurses, and therapistsare there to entertain the patient, while nature takes itscourse ." It synthesizes the prevailing view that once astroke patient arrives at a rehabilitation hospital there islittle that can be done to impact outcome . If nature werethe principal factor determining the stroke patient's out-come, the use of robots as "rehabilitators" t would be fun-damentally flawed . Therefore, the critical question weaddressed in the last 5 years was whether sensorimotortherapy influences brain recovery. In this paper we willsummarize our results from over 2,000 hours of robot-aided therapy with 76 stroke patients.

StrokeStroke rehabilitation is labor-intensive, usually rely-

ing on one-on-one, manual interactions with therapists.The demand for physical and occupational therapy forstroke survivors is expected to increase because improve-ments in medicine and health care will continue toincrease the life expectancy of the population, and theincidence of stroke is more prevalent among older adults.In fact, the relative incidence of stroke doubles withevery decade after age 55 and the U .S. demographic pat-terns compound the problem ; the leading edge of the"baby boom" will be 55 in a few years . Worldwide, the

I

641

KREBS et al . Robot-aided neuro-rehabilitation

r

World Health Organization (WHO) is predicting that thepopulation over 65 years old will increase by 88 percentin the coming years . New treatments for stroke are beingdeveloped but their impact on the need for other forms oftherapy is unclear. For example, as new pharmaceuticalagents for neuro-protection (e .g ., nerve growth factors,better receptor blockers, antioxidants, anti-inflammatoryagents, and blood clot dissolving agents) come into wide-spread use, the percentage of people surviving a strokemay increase, but the percentage of stroke victims poten-tially requiring rehabilitation may increase as well.

The effects of stroke can be devastating, resulting indeficits of cognitive, affective, sensory, and motor func-tions . Motor deficits persist chronically in about one-halfof stroke survivors (2) . Damage to neural areas responsi-ble for controlling movement and concomitant disuse andpersistent abnormal posture of the impaired limb cause ahost of centrally and peripherally based sensory andmotor impairments . Common impairments are decreasedpassive range of motion, weakness (3), hyperactivereflexes (4), and incoordination, manifest in part as aninability to independently co-activate muscles (5) . Thebiological processes that underlie recovery from neuro-logical injury remain a topic of intensive research . Aprominent theme of current neuroscience research intothe sequelae of brain injury posits that activity-dependentplasticity underlies neuro-recovery. If that is the case,there is good reason to believe that neurological changesthat may underlie recovery are facilitated by the standardpractice of providing targeted sensorimotor activity.

Stroke patients commonly experience some sponta-neous recovery, but are also treated with extensive phys-ical and occupational therapy. Because the variability ofbrain injury following stroke is enormous, in many casesit is unknown which therapies best promote recovery ; andbecause of the subjective nature of patient evaluation it isdifficult to monitor treatment effects precisely . Differentstudies have reported positive outcomes with severalapproaches, including repetitive passive exercises (6),forced use of the paretic limb by restraining the con-tralateral limb (7-9), increased amounts of therapyincluding external manipulation (10,11), and biofeedback(12) . On the other hand, comparative studies have gener-ally shown little difference among different therapeutictechniques (13,14) . In this paper, we will present one ofthese therapeutic approaches (our approach) : robot-aidedneuro-rehabilitation, which takes advantage of roboticsand information technology to enhance neuro-rehabilita-tion (Figure 1) .

Figure 1.A recovering stroke patient receiving upper extremity robot-aidedneuro-rehabilitation therapy.

Back-driveable Robot : MIT-MANUSTo determine if exercise therapy influences plas-

ticity and recovery of the brain following a stroke, weneeded a tool that would allow us to control the amountof therapy delivered to a patient, while objectivelymeasuring the patient's performance . Robotics can pro-vide this tool . Yet, the requirements for rehabilitation ofneurologically impaired patients impose unique con-straints on robot design . The first requirement for inter-acting safely with humans (impaired or otherwise) isthat the machine should be capable of gentle, compliantbehavior. In engineering terms, it should be highly"back-driveable" (equivalently, it should have lowintrinsic endpoint mechanical impedance) . Thisrequirement is difficult to satisfy with a commercialrobot . While a commercial robot could exert forces onthe patient's limbs, it shares current limitations of theindustrial robot technology . Because of a typicalrobot's electromechanical design and control architec-ture, it is intrinsically a position-controlled machineand does not yield easily under the action of externalforces. Active force feedback is needed to make therobot respond to the patients' actions, but that approachcannot (15) produce the "back-driveability" (lowmechanical impedance) required to move smoothly andrapidly to comply with the patients' actions.

The second requirement derives from special char-acteristics of neurologically impaired patients, who

642


often present abnormally low or (occasionally) highmuscle tone . As a result, for abnormally low muscletone, modest forces applied to the limbs can result inexcessive relative motion of limb segments . Likewise,excessive muscle tone might misguide the clinician toapply very large robot forces to obtain the desiredmotion of limb segments . The problem is particularlyacute for the shoulder where even the mild forcesapplied in standard therapy can injure the joint ifapplied improperly . Thus a neuro-rehabilitation robotcapable of activating arm and hand motion must also beable to control the forces exerted on the shoulder.Furthermore it must do so while guaranteeing stablebehavior despite almost complete ignorance of thedynamics of the neurologically impaired patient . Thatrequirement can be satisfied by ensuring the robotexhibits passive impedance; hence impedance controlis optimal.

To address the limitation of commercial robots, in1989 we started to specifically design and build a novellow-impedance robot for clinical neurological applica-tions capable of interacting safely and gently withhumans: MIT-MANUS (16, Figure 2) . Unlike mostindustrial robots, MIT-MANUS is configured for safe,stable, and compliant operation in close physical con-tact with humans. This is achieved using impedancecontrol, a key feature of the robot control system thatmodulates the way the robot reacts to mechanical per-turbation from a patient or clinician and ensures a gen-tle compliant behavior . Hogan (17) introducedimpedance control, and it has been extensively adoptedby other robotics researchers, especially those con-cerned with human-machine interaction . MIT-MANUScan move, guide, or perturb the movement of a sub-ject's or patient's upper limb, and can record motionsand mechanical quantities such as the position, veloci-ty, and forces applied . An overview of the robot's maincharacteristics can be found elsewhere (18) . This robothas been used daily for over 5 years with over 100stroke patients at the Burke Rehabilitation Hospital(White Plains, NY) . A second unit began operation inthe fall of 1999 at the Spaulding RehabilitationHospital (Boston, MA), while a third unit is scheduledto begin operation shortly at the Burke RehabilitationHospital and a fourth unit at the Baltimore VA Hospital(Baltimore, MD) . Our expectation is that these fourunits will substantially speed up the rate of researchand permit us to significantly impact the way rehabili-tation medicine is practiced .

Portable Robot

Figure 2.Exploded view of two-degree-of-freedom robot module.

Evidence-based Rehabilitation : Robot-aided Neuro .rehabilitation Benefits

Evidence-based rehabilitation might sound like a man-aged-care "buzzword" to justify further pruning of rehabil-itation expenses. Yet compassion alone cannot drive ahigh-quality efficient rehabilitation delivery system.Scientific evidence must qualify and quantify decisions thataffect patient care . A common language that clinicians canunderstand is needed; therefore, we opted to make use ofthe existing standard clinical assessment scales to presentadmissible evidence to the clinician that indeed, sensorimo-tor training influences brain recovery.

Robot-aided TherapySeventy-six sequential hemiparetic patients were

enrolled from 1995 to early 1999. Patients were admitted tothe same hospital wards and assigned to the same team ofrehabilitation professionals . They were enrolled in either arobot-aided sensorimotor therapy group (RT, N=40) or in agroup receiving standard therapy plus "sham" robot-aidedtherapy (ST, N=36) . Both groups are described in detailelsewhere (18-21). Patients and clinicians were blinded tothe treatment group (a double-blinded study) . Both groupsreceived conventional therapy ; the RT group received anadditional 4 to 5 hours per week of robot-aided therapy con-sisting of peripheral manipulation of the impaired shoulderand elbow correlated with audio-visual stimuli, while theST group had an hour of weekly robot exposure .

ti

643


The sensorimotor motor training for the RT group con-sisted of a set of "video games" and typically lasted for sixweeks. Patients were required to move the robot end-effec-tor according to the game's goals (Figure 3) . If the patientcould not perform the task, the robot assisted and guided the

patients hand. The robot was controlled by an impedance

controller, which produced a constant isotropic end-pointstiffness and damping . Coupled to our highly back-drive-able design, the stability of this controller is extremelyrobust to the uncertainties due to physical contact (22,23).The training for the ST group was similar to the RT group.Half of the one hour session consisted of playing the videogames with the unimpaired arm and half the session withthe impaired arm while the robot passively supported thearm and provided the video-game visual feedback (positionfeedback) . If the patient could not perform the task, he/sheused the unimpaired arm to assist the impaired arm andcomplete the game (self-ranging), or the clinician assisted.

Figure 3.Robot-aided neuro-rehabilitation task. Targets were arranged so thatdiagonal paths required predominantly elbow or shoulder motions, whilevertical, horizontal or curved paths (circle) required coordination of both.

A standard assessment procedure was used everyother week to assess all patients during rehabilitation andduring the recall post-hospital discharge (robot-aidedtherapy group and control group) . This assessment wasalways performed by the same "blinded" rehabilitationprofessional . Each patient's motor function was assessedby standard procedures including : the FunctionalIndependence (FIM), the upper limb subsection of theFugl-Meyer (F-M), Motor Power for shoulder and elbow(MP), Motor Status Score for shoulder and elbow (MS1),and Motor Status Score for wrist and fingers (MS2).

In-patient Benefits

Table 1 presents the composite results of two trialswith 76 patients (initial study with 20 patients, see refer-

Table 1.Change during acute rehabilitation (76 patients)

Group F-M MP MS1 MS2

RT (40 patients)Al

9 .2 .5±1 .36Al*

3 .99±0 .43

Al*

8 .15±0 .79Al

4 .16±1 .1.6

ST (36 patients) 7.1±1 .20 2 .0±0 .32 3 .42±0.62 2.64±0.78

Experimental (RT) vs . Control (ST) Group--Al : score change from rehabilita-

tion hospital admission to discharge ; one-way t-test that RT>ST with p<0 .05

for statistical significance (*).

ences 18—20; replication study with 56 patients, see ref-erence 21).

The F-M and MS2 show no statistically significantdifference between groups . The MS 1 and MP for shoulderand elbow show a statistically significant improvement : theexperimental group responding to therapy with about twicethe score improvement of the control group over a compa-rable period . Our replication study confirmed and strength-ened our initial pilot study (19) . Note that results for theadditional 56 patients were of the same order of magnitudeas the ones in the initial study (21) . The difference betweenthe results with the different measures may be in part due todifferences in the resolution of these instruments and whatthey measure. For example, the F-M measures motorbehavior most sensitively from the acute injury (a timewhen the affected limb may be flaccid) to a point when thelimb is developing tone and reflex changes and synergy . Forthat reason we adopted the MSS scale, which was devel-oped at the Burke Rehabilitation Hospital for a previousstudy. The MSS scale takes the F-M measures and focusesthe motor analysis on the movement about the shoulder,elbow, wrist, and fingers.

The MSS for elbow and shoulder (MS1) consists ofa sum of scores (0 to 2) given to 10 shoulder movementsand 4 elbow/forearm movements . The MSS for wrist andfingers (MS2) consists of a sum of grades for three wristmovements and 12 hand movements (19—21) . The MotorPower Score includes the standard six-point scale assess-ing muscle power in biceps, triceps, and anterior and lat-eral deltoid muscles, and is sometimes referred to as theOxford Scale (21) . A notable feature of these results isthat although the MS1 is capable of detecting a signifi-cant advantage of robot therapy for shoulder and elbow,the MS2 for wrist and fingers shows no significant dif-ference between experimental and control groups . As it isunlikely that the lack of statistical significance is due toinadequate resolution of this measure, the result suggeststhat the benefit of robot-administered sensorimotor train-ing (and perhaps human-administered sensorimotor train-

644


ing, too) is specific to the muscle groups or limb seg-ments exercised and does not generalize broadly.

Long-term Benefits

To test whether motor advantages conferred on therobot-trained group would persist, we recalled patientsenrolled in the first pilot clinical trial (20 patients) 3 yearsafter hospital discharge . If the improved outcome was notsustainable, one might conclude that manipulation of theimpaired limb influenced the rate of recovery during theinpatient post-stroke phase, but not at the "final" plateau.Twelve of these 20 inpatients were successfully recalledand evaluated by the same "blinded" therapist (of theremaining 8 patients, 4 could not be located, 1 had died, and3 had a second stroke or other medical complications) . Sixpatients in the RT and in the ST groups were comparable ingender distribution, lesion size (RT=53 .8±1113 .3±59, ST:960±81 days) . There was no control over patients' activitiesafter hospital discharge. Results are shown in Table 2 andwere described elsewhere (20,25) . Summarizing, theimproved outcome during inpatient rehabilitation was sus-tained after 3 years post discharge ; and the improvementwas again confined to the muscle groups trained in therobot-aided therapy, i .e., shoulder and elbow.

This data should be interpreted with care due to thesmall number of subjects . Nevertheless, comparing theoverall recovery (between admission and 3 years afterdischarge) the MSS for shoulder and elbow (which werethe focus of robot training) of the experimental groupimproved twice as much as that of the control group(MS1 - A3 score), whereas the MSS of wrist and fingers(which were not robot trained) improved by essentiallythe same amount for both groups (MS2 - A3 score) . Notealso that both groups had comparable improvementbetween hospital discharge and 3-year recall (periodwithout robot-aided therapy, A2 score) . These results cor-roborate our inpatient studies (Table 1), indicating thatthe benefits of robot training are specific to the musclegroups or limb segments exercised . Furthermore, it is

Table 2.Change during acute rehabilitation and follow-up (12 patients)

F-M (out of 66)

MP (out of 20)

striking that eight out of twelve patients who were suc-cessfully recalled continued to improve substantially inthe period following discharge (RT and ST subjects).

If this finding is corroborated in the recall of thereplication study group (56 patients recall in progress),it would challenge the common perception that patientsstop improving after about 11 weeks post-stroke (e .g .,

The Copenhagen Stroke Study, see reference 26) . Ourresults suggest the possibility that the use of unsuitablycoarse scales may misjudge patients' potential and thatthere may be an opportunity to further improve the motorrecovery of some stroke patients by continuing therapy inthe outpatient phase.

Neuro-recovery Time History Changes with LesionTerritory

The fundamental mechanisms underlying neuro-recovery are understood poorly at best. The so-called"activity-dependent plasticity" that is posited to driverecovery may be due, in part, to the unmasking of pre-existing connections, focal synaptic changes, or neosy-naptogenesis the growth of new connections.Experimental support for this idea derives primarily frommeasurements of synaptic branching and cortical thick-ness in rats raised in enriched environments and deprivedenvironments (e .g ., 27—31) and in monkeys recoveringfrom ischemic injury (e .g ., 32) . One challenge is tounderstand whether the neurobiologic mechanism for thechanged motor behavior is based on reorganization ofnormal cortex that surrounds the injury, or of more distantsupplemental motor circuits (in the supplemental motorarea, the basal ganglia, or cerebellum), or of the unaffect-ed hemisphere (33-35) . Another (and probably related)mechanism involves assumption of lost function by adja-cent areas of undamaged brain tissue . Reorganization ofcortical maps has been demonstrated in the motor system(36,37), sensory system (38,39), visual system (40), andauditory system (41) . A further mechanism of recovery offunction post stroke involves the homologous regions of

MS1 (out of40)

MS2 (out of 42)

Al

A2

A3

Al*

A2

A3

RT

15.3

5 .0

20 .3

4 .5

4.6

9 .1

ST

8 .0

12 .3

20 .3

1 .6

3 .5

5 .1

Al*

A2

A3

Al

A2

A3

12 .0

9 .4

21 .4

8 .2

8 .3

16 .4

-1 .0

10 .2

9 .2

3 .7

8 .0

11 .7

Experimental (RT) vs . Control (ST) Group - Al : score change from rehabilitation hospital admission to discharge ; A2 : score change from discharge to follow up;

03 : score change from admission to follow up; one-way t-test that RT>ST with p<0 .05 for statistical significance (*) .

1

645


the unaffected contralateral cerebral hemisphere substi-tuting for the infarcted brain tissue (42-44) . A mechanismby which motor function may be controlled by the unaf-fected ipsilateral hemisphere may be through the 25 per-cent (45) of pyramidal tract fibers that are uncrossed.

Activity-dependent cortical plasticity has beenfound to accompany motor learning, and rehabilitationand training after injury have also been reported to influ-ence the pattern of reorganization . These findings sug-gested that at least some aspects of the recovery of motorbehavior in stroke patients should exhibit characteristicsnormally associated with motor learning . Our own resultsto date are consistent with this hypothesis . To the extentthat sensorimotor training facilitates a process akin tomotor learning, its benefits should be specific to the limbsegments involved in the training, and that is what wehave observed. Whereas the arm and forearm, which arethe focus of our robot training, show significant reductionof impairment, the wrist and fingers, which are not afocus of this training, show no significant benefit of robottraining.

Given this state of knowledge, the best predictors ofoutcome remain a topic of research . Intuitively one mightexpect that larger lesions would lead to poorer outcomes,but Miyai and colleagues showed that while lesion size isan important variable in predicting stroke outcome, lesionterritory might be even more important . They showed thatpatients with stroke confined to basal ganglia (CS) withsmaller lesions have diminished response to rehabilitationefforts compared to patients with much larger lesions thatinvolve both the basal ganglia and cortical territories(CS + ; reference 46) . They suggested that isolated basalganglia strokes might cause persistent corticothalamic-basal ganglia interactions that are dysfunctional andimpede recovery. At face value, it would seem that senso-rimotor therapy would have little benefit for this (CS)group. However, we must go beyond the inpatient phaseand track the whole process to fully understand theprocess of neuro-recovery .

For the patients recalled in the follow-up describedabove, CT scans showed six pure subcortical and six sub-cortical plus cortical lesions . The comparison of outcomefor 5 patients with corpus striatum lesions (CS) versus 6

patients with corpus striatum plus cortex (CS+) is shownin Table 3 and Figure 4 (one patient with rapid motorrecovery after an isolated thalamic injury was excluded;see reference 25) . These patients had comparable demo-graphics and were evaluated by the same therapist on hos-pital admission (19 days + 2 post-stroke), discharge (33days + 3 later), and follow-up (1,002 days + 56 post-dis-charge) . As in the study by Miyai et al ., the CS group hadsmaller lesion size (CS=13 .3±3 .9 cm3, CS+=95 .1±25.2cm3, p<0 .05).

Our results are consistent with the observation ofMiyai et al . (46) for the time period that covered the acuterehabilitation phase . Note in Table 3 that the CS+ groupoutperformed the CS group during inpatient rehabilita-tion. However, the follow-up reinforced the old clinicaltruism that anatomy is destiny : It revealed that patientswith smaller lesions eventually fare better. Specifically,note that the CS group outperformed the CS+ groupbetween discharge to follow-up. In fact, the CS groupoutcome is far superior at follow-up . This clinical resultbears some resemblance to the important problem ofdelayed neuronal degeneration . There are several animalmodels in which initial injury in the basal ganglia areaccompanied by neuronal degeneration in neurons distantfrom the initial injury and occurring over longer periodsof time (e .g ., 47-49) . This further reinforces the impres-sion that motor recovery during inpatient rehabilitationmay not be complete.

Understanding motor recovery will require longitu-dinal studies beyond the inpatient period. Otherwise wemight wrongly conclude that therapy should be discon-tinued to patients with stroke confined to the CS territo-ries, because they appear not to respond. In fact, the CSgroup in our study outperformed the CS + following hos-pital discharge . For cost reasons, extending inpatient

Table 3.Change during acute rehabilitation and follow-up

Group F-M (out of 66) MP (out of 20) MS1 (out of 40) MS2 (out of 42)

Al A2* A3* Al A2 A3 Al A2* A3* Al A2* A3*CS 9 .3 25 .0 34.3 2.1 6 .1 8 .2 1 .0 16 .0 17 .0 10 .0 14 .5 24 .5CS+ 10 .7 -1 .3 9 .4 4 .3 2 .8 7 .1 7 .7 4 .2 11 .9 3 .3 3 .2 6 .5

Lesion Site Classification--Al : score change from rehabilitation hospital admission to discharge ; A2 : score change from discharge to follow up ; A3 : score changefrom admission to follow up ; one-way t-test CS>CS+ with p<0 .05 for statistical significance (*).

646


cm , ^g ° mmm

Figure 4.

Change during acute rehabilitation and follow-up : Lesion site classification.

rehabilitation is unlikely to be practical . However, therobotic neuro-rehabilitation technology might facilitatethe extension of therapy for neuro-rehabilitation farbeyond the inpatient phase . For example, it may enablerobot-assisted self-therapy in a home setting . Connectionto the Internet may permit low-cost periodic evaluationby appropriate clinical personnel, or interactions withother patients at similar stages of recovery.

Robot-aided Therapy Interacts with Lesion AnatomyOur repeated finding that patients who received

robotic training enjoyed significantly improved motoroutcome is encouraging. However to assess the truepotential of robot-aided neuro-rehabilitation, we need tounderstand the biological basis of recovery. As outlinedabove, our initial approach is to examine the relationbetween lesion anatomy and the effects of therapy . Doeslesion territory determine functional outcome? More suc-cinctly, is anatomy destiny? Lesion anatomy is clearly acritical factor—for instance, the absence of any lesionwould surely obviate the need for therapy—but perhapsthe more definitive question is whether lesion anatomydictates the effectiveness of robot therapy.

Consider patients with middle cerebral arterylesions (MCA) involving the pre-motor area (PMC), orsparing it . Table 4 and the top row of Figure 5 show the

motor power scores of 33 of these patients (14 patientswith lesion involving the PMC and 19 patients withspared PMC). Reclassifying our patients according towhether the PMC and the aobcor(icu!effereutafrom thePMC were damaged suggested that those patients withspared PMC were better at the first and final evaluationon the MP scale . Similar results were obtained for theother scale of impairment of the shoulder and upper armmuscles, the MS1 (data not shown). In particular, theydemonstrated the facilitator effect PMC sparing had onshoulder and upper-arm motor function.

These results are in agreement with recent studiesindicating that patients for whom the PMC was spared(sPMC) recover significantly better than the ones withlesion foci that includes the PMC (50) . Patients withcortical and subcortical damage of comparable volumehad different functional outcome depending on whetherthe PMC was damaged . Results from other invewdga-torouoiug a variety of functional cerebral imaging ieob-niqueabave also pointed to the PMC as a crucial regionof activation during motor recovery (51 `52). Theemerging importance of the PMC in motor recoveryfinds additional support in the in vivo experimental lit-erature.

Because we are also interested in the effect ofrobot training, these data were further analyzed consid-

647


Table 4.Motor power scores

(out of 20)

PMC (14 patients)

sPMC (19 patients)

MP-admission

1 .1910.83

3 .95±1 .10

MP-discharge

3 .66±0.86

7 .24±1 .02

Motor power scores at admission and discharge of patients with MCA lesionincluding or excluding the premotor territories (PMC or sPMC, respectively;ANOVA for groups being different; p-value 0 .0027).

Figure 5.Positive interaction robot training and PMC status.

ering such training as an additional independent vari-able (bottom row of Figure 5). While the sample sizewas small, there was a clear trend in both the PMC-spared and -damaged groups for the robot training topositively impact the outcome, i .e ., patients in therobot-aided sensorimotor training program, indepen-dent of PMC status, improved more than the subjects inthe control group . This finding further suggests thatwhile lesion anatomy is clearly the critical factor, itdoes not dictate the effectiveness of robot therapy.

One must take the above results with the appropriatecaveats : Anatomy may not be destiny, but it appears to bea close relative . Understanding motor recovery will allowus to best challenge fate . For example, in the 56-patientreplication study, a histogram of the number of patientsper lesion volume (bins of 25 cm 3) suggested a bimodal

distribution, indicating two distinct classes of patients:one with lesion volume smaller than 100 cm3 (N=42) andanother with lesions larger than 100 cm3 (N=14). Whilean analysis of whether the differences in motor outcomemight result from lesion volume alone were unrevealing,of those in the group of 42 patients with smaller lesionvolume, the ones exposed to the robot sensorimotor train-ing outranked the remaining ones not exposed to this kindof focused exercise (MP and MS1 scales).

Robot-based MeasurementsA common language facilitates communication.

Therefore in the previous sections, we refrained fromusing any measurement but standard clinical scales tomake the point to the clinical community that ourresults to date indicate that exercise therapy has a gen-uine positive effect on brain recovery following astroke. Yet while we might be able to impact the neuro-recovery process, we have only touched the tip of theiceberg and harder questions lie ahead. Of particularimportance is how to tailor and optimize therapy to theparticular patient's need. While clinical scales alonecan help us trace the big picture, their coarse naturerequires extensive and time-consuming trials and ontop of that they fail to show us details important foroptimizing therapy. Our goal in this section is twofold:First, we want to emphasize that robotic technologyoffers the potential benefit of new measurements—anddeeper insight—into the process of recovery from neu-rological injury; and second, we want to emphasize theimportance of back-driveable robots not only for deliv-ering therapy, but also for measurement.

To that end we will revisit the example depicted inFigure 6 (18), which shows the movements of a recov-ering stroke patient attempting to draw a circle in a hor-izontal plane . At week 6 post-stroke, the patient had justregained the control of the elbow extension and can per-form the task; the movement is rather uncoordinated andjerky. By week 11 the path is more nearly circular andthe speed fluctuations have diminished and it wouldappear eminently reasonable to conclude that the move-ment at week 11 is superior to the movement at week 6.Nevertheless, the standard clinical measures failed toindicate any difference ; in fact, the most sensitive ofthese clinical measures for shoulder and elbow, theMotor Status Score for shoulder and elbow (MS1),indicated the same score value.

Another example of the insight that may be acquiredfrom robot data is presented in Figure 7, which shows

648

6

M31=16

Figure 6.Movements of a recovering stroke patient attempting to draw a circle in a horizontal plane . The left column shows a plan view of the patient'shand path . The right column shows the tangential speed of the hand along the path plotted against time. Note that the scored value for MS 1 (shoul-der and elbow) was the same.


0 .2

J

E -02-

-0 .2-0 .1 0 0 .111 .2C0 E

a 0 .4V)

0 .2

-02

0-0 .2-0.1 0 0 .1 0 .2

0 2 4 6

x-direction (m)

time (sec)

P16 Initial and final

position P2

0 .25

0 .2

0 .15

n

0 .15

0 .1iii-

0 .05

-0 .1 -0 .05

0

0 .05 0 .1

0 .15

x-direction (m)0

10

20 3047 (cc

C s0 .15

E 0 .1

-0 .1

0 .25

0 .2

0 .15

0 .1

0 .05

0-0 .1 -0 .05 0 0 .05 0.1 0.15

x-direction (rn)

30

time (tie(0 2010

Figure 7.Reaching movements made by patients with corpus striatum lesion—CS (8 .9 cm3 ) and corpus striatum plus cortex—CS+ (109 .9 cm3) lesions.

The left column shows a plan view of the patients' hand path attempting a point-to-point movement . The right column shows hand speed .

649


representative reaching movements made by two patientswith different brain lesions . The left column shows a planview of the patient's hand path when attempting to movefrom one position to another, as indicated . The right col-umn shows the tangential speed of the hand along thepath plotted against time . The top two rows were record-ed from a patient with a single ischemic infarct in themotor cortex . The bottom two rows were recorded from apatient with a single ischemic infarct in the basal gangliaoutside the internal capsule . Comparing the two patients,note that the patient with the basal ganglia lesion appearsto move exceptionally slowly ; however, the hand path isgenerally well aimed towards the target position . In con-trast, the patient with the motor cortical lesion makes aseries of moves with much higher peak speeds (approach-ing the movement speeds of unimpaired subjects) buteach of these movements appears to be poorly aimed atthe target position (53) . The cortical patient's mis-aimingappears to be consistent with the observation (e.g ., 54)that activities of populations of motor cortical neuronsare correlated with the intended direction of reachingmovements.

Observations such as these are the basis for ourbelief that robot-based instrumentation can provide finer-grained, higher-resolution measures of recovery. In fact,borrowing from a century-old conjecture (55), we havebeen developing techniques for movement analysis basedon the concept of segmentation of apparently continuousmovement (submovements) . By studying the kinematicsof movements made by patients with neurological injury,we may have discovered the kinematic profile of a prim-itive unit action (56) . This temporal motor primitive is inline with Woodworth's conjecture that a repertoire ofmovement primitives constitutes the fundamental build-ing blocks of complex motions . The fact that we haveidentified a precise mathematical characterization of sub-movement kinematics provides the key information thatmakes it possible to de-convolve continuous movementsobjectively and reliably into their component submove-ments . That is, it changes a "hard inverse problem" into afiltering problem.

These examples illustrate the potential benefit of bet-ter measurements that afford deeper insight into the processof neuro-recovery. Furthermore, they illustrate the impor-tance of back-driveable robots not only for delivering ther-apy, but also for concomitantly measuring patient behavior.As with the design of any instrument, we must ensure thatthe measurement process does not corrupt the quantity to bemeasured . In this case, to avoid suppressing dynamic

details of patient movement the robot should not encumberthe patient; it should have minimal mechanical impedance(ideally zero) . This is not a trivial requirement . To be spe-

cific, the data shown in Figures 6 and 7 and also shown inKrebs et al . (56) could not have been obtained using a typ-ical commercial robot, even with active force feedback,because that approach cannot produce the "back-driveabil-ity" required to move smoothly and rapidly to comply withthe patient's actions.

Increase in ProductivityOne common misperception is that robot therapy

would ultimately replace human-administered therapy . Tothose afraid of this kind of technology, we offer a quote byIsaac Asimov "I do not fear computers, I fear the lack ofthem." Just as word processors opened the door to marked-ly increased efficiencies for office workers, robot assistantspromise the same for rehabilitation clinicians . We envisionthe role of the therapist evolving from delivering repetitivelabor-intensive manual treatment to a more supervisorydecision-making capacity. Productivity will soar if the con-cept of robot-aided classrooms lives up to its promise, i .e .,delivering individual therapy to more than one patient at atime without compromising quality or dosage . Figure 8shows a composite illustrating the concept of a classroom inwhich a therapist oversees four patients either directly orvia a robot-aided workstation, as each patient interacts witha robot.

In keeping with our caution against raising prematureand/or unrealistic clinical expectations, our initial objectiveis to establish that robot therapy can be equivalent in quali-ty to traditional methods, but it will deliver therapy atreduced cost and with increased institutional controls . Wetherefore are working to identify patterns of use for thedevice, together with all underlying reimbursement mecha-nisms, to confirm that from an efficiency-minded clinic'spoint of view, robot therapy confers the benefits of desirabletechnology innovations : the accomplishment of everydaytasks more efficiently, and with increased precision . Towardthat end, we are identifying how Burke's patient censustranslates into daily hours of usage of the device, as well asthe nature of those hours (e .g ., diagnostic versus therapydelivery) . We are also identifying the sources of reimburse-ment for robot therapy—Medicare, private payor, fee forservice—and how those sources may affect the daily hoursof use of the device . As an ancillary issue, we are identify-ing patients who may be amenable to commencing robottherapy at the clinic, and then continuing with a home-based therapy .

650


Figure 8,Robot-aided classroom.

CNCLUSIN

At the onset of this research we had to confrontsquarely (and solve!) a critical question : If anatomy is des-tiny, can we influence it? This research question has beenthe focus of our concern over the last 5 years, and in pre-senting a few of our clinical results of over 2,000 hours ofrobot-aided therapy with 76 stroke patients, we have barelyscratched the surface . Nevertheless, all of the indications todate suggest that robot-aided sensorimotor training doeshave a genuinely positive effect on reduction of impairmentand the reorganization of the adult brain . Our results are inagreement with one of the prominent themes of current neu-roscience research into the sequelae of brain injury or trau-

ma, which posits that activity-dependent plasticity underliesneuro-recovery. In other words, there is good reason tobelieve that neurological changes that may underlie recov-ery are facilitated by the standard practice of providing tar-geted sensorimotor activity, and that this can beaccomplished using robot technology.

Nevertheless, the fundamental mechanisms underly-ing neuro-recovery following a stroke remain poorlyunderstood . Considerable further study is needed to under-stand the biological basis of recovery and to optimize treat-ment to meet a particular patient's needs . To that end, weneed the appropriate "microscope ." Back-driveable robotsare the key enabling technology that allow us concomi-tantly to control the amount of therapy delivered to a

651


patient while objectively measuring the patient's perfor-mance. We also plan to use present technology to establishthe practicality and economic feasibility of clinician-super-vised, robot-administered therapy, including classroomtherapy. We feel quite optimistic that the march of progresswill accelerate substantially in the near future and allow usto transfer this technology from the research realm to theeveryday treatment of stroke survivors.

REFERENCES

1. Krugman P. The age of diminished expectations . Cambridge : TheMIT Press; 1994.

2. Gresham GE, Phillips TF, Wolf PA, McNamara PM, Kannel WB,Dawber TR . Epidemiologic profile of long term disability instroke: The Framingham Study . Arch Phys Med Rehabil1979 ;60 :487-91.

3. Gandevia SC . Strength changes in hemiparesis : Measurementsand mechanisms . In : Thilmann AF, Burke DJ, Rymer WZ, editors.Spasticity : Mechanisms and management . Berlin : Springer-Verlag ; 1993 . p . 111-22.

4. Thilmann AF, Burke DJ, Rymer WZ, editors . Spasticity:Mechanisms and management . Berlin : Springer-Verlag ; 1993.

5. Dewald JPA, Pope PS, Given JD, Buchanan TS, Rymer WZ.Abnormal muscle coactivation patterns during isometric torquegeneration at the elbow and shoulder in hemiparetic subjects.Brain 1995 ;118 :495-510.

6. Butefisch C, Hummelsheim H, Denzler P, Maurtiz KH . Repetitivetraining of isolated movements improves the outcome of motorrehabilitation of the centrally paretic hand. J Neurol Sci1995 ;130 :59-68.

7. Wolf SL, LeCraw DE, Barton LA, Jann BB . Forced use of hemi-plegic upper extremities to reverse the effect of learned nonuseamong chronic stroke and head-injured patients . Exp Neurol1989;104(2) :125.

8. Taub E, Miller NE, Novack TA, Cook EW 3d, Fleming WC,Nepomuceno CS, et al. Technique to improve chronic motordeficit after stroke . Arch Phys Med Rehabil 1993 ;74 :347-54.

9. Miltner WH, Bauder H, Sommer M, Dettmers C, Taub E. Effectsof constraint induced movement therapy on patients with chronicmotor deficits after stroke : A replication . Stroke 1999 ;30 :586-92.

10.Sunderland A, Tinson DJ, Bradley EL, Fletcher D, Langton HewerR, Wade DT. Enhanced physical therapy improves recovery of armfunction after stroke . A randomized controlled trial . J NeurolNeurosurg Psychiatry 1992 ;55 :530-5.

11.Dam M, Toni)) P, Casson S, Ermani M, Pizzolato G, Iaia V,Battistin L. The effects of long-term rehabilitation therapy on post-stroke hemiplegic patients . Stroke 1993 ;24(8) :1186-91.

12. Wolf SL, Binder-MacLeod SA . Electromyographic biofeedbackapplications to the hemiplegic patient: Changes in upper extremi-ty neuromuscular and functional status . Phys Ther 1983 ;63 :1 .393.

13. Logigian MK, Samuels MA, Falconer JF . Clinical exercise trial forstroke patients . Arch Phys Med Rehabil 1983 ;64 :364-7.

14. Dickstein R, Hocherman S, Pillar T, Shaham R. Stroke rehabilitation:Three exercise therapy approaches . Phys Ther 1986 ;66 :1233-8.

15. Lawrence DA. Impedance control stability properties in common

implementations . Proceedings of the IEEE InternationalConference on Robotics and Automation ; 1988 ; p. 1185-91.

16. Hogan N, Krebs HI, Sharon A, Charnnarong J. Interactive robotic

therapist . U .S . Patent #5,466,213, MIT; 1995.

17. Hogan N . Impedance control : An approach to manipulation . Part I,

part II, part III . ASME J Dynamic Systems, Measurement and

Control 1985 ;107 :1-24.18.Krebs HI, Hogan N, Aisen ML, Volpe BT . Robot-aided neuro-

rehabilitation, IEEE Trans Rehabil Eng 1998 ;6(1) :75-87.

19.Aisen ML, Krebs HI, McDowell F, Hogan N, Volpe N . The effectof robot assisted therapy and rehabilitative training on motorrecovery following stroke . Arch Neurol 1997 ;54 :443-6.

20. Volpe BT, Krebs HI, Hogan N, Edelstein L, Diels C, Aisen ML.Robot training enhanced motor outcome in patients with strokemaintained over 3 years . Neurology 1999 ;53 :1874-6.

21. Aisen ML, Sevilla D, Gibson G, Kutt H, Blau A, Edelstein L, et al.3,4-Diaminopyidine as a treatment for amyotrophic lateral sclero-sis . J Neurol Sci 1995 ;129 :21-4.

22. Volpe BT, Krebs HI, Hogan N, Edelstein L, Diels C, Aisen ML . Anovel approach to stroke rehabilitation : Robot-aided sensorimotorstimulation . Neurology 2000 ;54:1938-44.

23. Hogan N . Impedance control : An approach to manipulation.American Control Conference; 1984 ; p . 304-13.

24. Colgate JE, Hogan N . Robust control of dynamically interactingsystems . Int J Control 1988 ;48(1) :65-88.

25. Krebs HI, Hogan N, Volpe BT, Aisen ML, Edelstein L, Diels C.Robot-aided neuro-rehabilitation in stroke : Three year follow-up.IEEE - ICORR; 1999 ; p . 34-41.

26. Jorgensen HS, Nakayama H, Raaschou HO, Vive-Larsen J, StoierM, Olsen TS . Outcome and time course of recovery in stroke, PartI : Outcome, Part II : Time course of recovery, Copenhagen StrokeStudy. Arch Phys Med Rehabil 1995 ;76 :5 :399-412.

27. Diamond MC, Greer ER, York A, Lewis D, Barton T, Lin J . Ratcortical morphology following crowded-enriched living condi-tions . Exp Neurol 1987 ;96(2) :241-7.

28. Diamond MC, Ingham CA, Johnson RE, Bennett EL, RosenzweigMR. Effects of environment on morphology of rat cerebral cortexand hippocampus . J Neurobiol 1976 ;7(1) :75-85.

29. Diamond MC, Johnson RE, Protti AM, Ott C, Kajisa L . Plasticityin the 904-day-old male rat cerebral cortex . Exp Neurol1985 ;87(2) :309-17.

30. Diamond MC, Law F, Rhodes H, Lindner B, Rosenzweig MR,Krech D, Bennett EL . Increases in cortical depth and glia numbersin rats subjected to enriched environment. J Comp Neurol1966;128(1) :117-26.

31. Greer ER, Diamond MC, Murphy GM Jr . Increased branching ofbasal dendrites on pyramidal neurons in the occipital cortex ofhomozygous Brattleboro rats in standard and enriched environ-mental conditions : A Golgi study. Exp Neurol 1982 ;76(2) :254-62.

32.Nudo RJ, Wise BM, SiFuentes F, Milliken GW. Neural substratesfor the effects of rehabilitative training on motor recovery afterischemic infarct. Science 1996 ;272 :1791-4.

33. Jenkins WM, Merzenich MM . Reorganization of neocortical rep-resentations after brain injury. Prog Brain Res 1987 ;71 :249-66.

34. Aizawa H, Inase M, Mushiake H, Shima K, Tanji J.Reorganization of the supplementary motor areas after neocorticaldamage. Exp Brain Res 1994 ;84 :668-71 .

652


35. Jones TA, Shallert T. Use dependent growth after neocortical dam-age . J Neurosci 1994 ;14 :2140-52.

36. Asanuma C. Mapping movements within a moving motor map.Trends Neurosci 1991 ;14 :217-8.

37. Jacobs KM, Konoghue JP. Reshaping the cortical motor map byunmasking latent intracortical connections. Science 1991 ;251 :944-7.

38. Merzenich MM, RJ Nelson, Stryker MP, Cynader MS, SchoppmannA, Zook JM . Somatosensory cortical map changes following digitamputation in adult monkeys . J Comp Neurol 1984 ;224 :591-605.

39. Pons TP, Garraghty PE, Mishkin M . Lesion-induced plasticity inthe second somatosensory cortex of adult macaques . Proc NatlAcad Sci USA 1988 ;85 :5279-81.

40. Kaas JH, Krubitzer LA, Chino YM, Langston AL, Polley EH,Blair N . Reorganization of retinotopic cortical maps in adult mam-mals after lesions of the retina . Science 1990 ;248 :229-31.

41. King AJ, Moore DR. Plasticity of auditory maps in the brain.Trends Neurosci 1991 ;14 :21-7.

42. Fisher CM . Concerning the mechanism of recovery in strokehemiplegia . Can J Neurol Sci 1992 ;19 :57-63.

43. Glees P. Functional reorganization following hemispherectomy inman and after small experimental lesions in primates . In : Bach-y-Rita P, editor. Recovery of function: Theoretical considerations forbrain injury rehabilitation . Baltimore : University Park Press ; 1980.

44. Sabatini U, Toni D, Pantano P. Motor recovery after early braindamage . Stroke 1994 ;25 :514-7.

45. Nyberg-Hansen R, Rinvik E. Some comments on the pyramidaltract, with special reference to its individual variations in man.Acta Neurol Scand 1963 ;39 :1-30.

46. Miyai I, Blau AD, Reding MJ, Volpe BT . Patients with stroke con-fined to basal ganglia have diminished response to rehabilitationefforts . Neurology 1997 ;48 :95-101 .

47. Saji M, Endo Y, Miyanishi T, Volpe BT, Ohno K . Behavioral cor-relates transneuronal degeneration of substantia nigra reticulataneurons are reversed by ablation of subthalamic nucleus . BehavBrain Res 1997 ;84 :63-71.

48. Stefanis L, Burke RE . Transneuronal degeneration in substantianigra pars reticulata following striatal excitotoxic injury in adultrat: Time-course, distribution, and morphology of cell death.Neuroscience 1996 ;74(4) :997-1008.

49. Du C, Rong H, Csernansky CA, Hsu CY, Choi DW . Very delayedinfarction after mild focal cerebral ischemia . J Cereb Blood FlowMetab 1996 ;16 :195-201.

50. Pantano P, Formisano R, Ricci M, DiPiero V, Sabatini U, BarbantiP, et al . Prolonged muscular flaccidity after stroke : Morphologicaland functional brain alterations . Brain 1995 ;118 :1329-38.

51. Chollet F, DiPiero V, Wise RJS, Brooks DJ, Dolan RJ, FrackowiakRSJ. The functional anatomy of motor recovery after stroke inhumans : A study with positron emission tomography . Ann Neurol1991 ;29(1) :63-71.

52. Frackowiak RSJ . Plasticity and the human brain : Insights fromfunctional imaging . Neuroscientist 1996 ;2(6) :353-62.

53. Aisen ML, Krebs HI, Hogan N, Volpe N . Lesion location andspeed-accuracy tradeoff in stroke patients . Proceedings of theAmerican Academy of Neurology ; 1998.

54. Georgopoulos AP, Kalaska JF, Crutcher MD, Caminiti R, MasseyJT. The representation of movement direction in the motor cortex:Single cell and population studies . In: Dynamic Aspects ofNeocortical Function . New York : John Wiley & Sons ; 1984.

55. Woodworth RS . Ph .D . thesis, Columbia University ; 1899.56. Krebs HI, Hogan N, Aisen ML, Volpe BT. Quantization of contin-

uous arm movements in humans with brain injury . Proc Natl AcadSci USA 1999 ;96 :4645-9 .

Documents

Increasing productivity and quality of care: Robot-aided ...KREBS et al. Robot-aided neuro-rehabilitation r World Health Organization (WHO) is predicting that the population over 65